Classification of urban tree species based on seasonal traits using time series of multispectral satellite data

Trees are considered as an important component in urban ecosystems, providing both aesthetic views and many ecological benefits. Therefore, accurate classification of tree species in urban environments has become an issue of current interest. The purpose of the presented study is to derive a methodological framework to analyse time series of intra-annual multispectral satellite data for accurate classification and mapping of urban tree species and mixed groups using their extracted seasonal traits. Conventionally, it can be divided into two stages. The first stage is devoted to the extraction of urban tree species seasonal traits obtained from SVI time series and refinement of training data sets. The second one is concerned urban tree classification itself using any single multispectral from the acquired time series of satellite imagery. The proposed methodology was applied to classify tree species within Pushkin Park and the park of Kyiv Polytechnic Institute. T he main output data product representing map of the tree species and their mixed groups for 12 recognized tree species. This research demonstrates that seasonal traits extraction from time series of intra-annual multispectral satellite data is very promising for classification of urban tree species and mixed groups.